Documentation Agent Tutorial

🟢 Beginner • 30-60 minutes • AgentFlow + Knowledge Base

Build an AI documentation agent using Axellero Studio's AgentFlow Builder. This tutorial demonstrates core platform concepts through implementation of an agent that answers questions about documentation using RAG and LLM integrations.

What You'll Build

This tutorial creates a functional AI agent that:

• Connects to LLM providers through Axellero Studio's integration interface

• Accesses uploaded documentation via knowledge base with RAG

• Operates through the agentic loop for multi-step reasoning

• Deploys to production with version control and security settings

Prerequisites

• Axellero Studio account - With access to AgentFlow Builder and integrations
• LLM provider credentials - API keys for OpenAI (recommended) or Anthropic
• Platform familiarity - Completion of platform introduction or equivalent experience
• Axellero application - You have created Axellero App for this tutorial

Agent Architecture

┌─────────────────────────────────────────────────┐
│          AgentFlow: Documentation Expert        │
├─────────────────────────────────────────────────┤
│  ┌─────────────┐  ┌─────────────┐  ┌──────────┐ │
│  │   LLM       │  │ Knowledge   │  │  Tools   │ │
│  │ Connection  │  │    Base     │  │   Node   │ │
│  │   Node      │  │    Node     │  │          │ │
│  │ (OpenAI)    │  │ (RAG Docs)  │  │JavaScript│ │
│  └─────┬───────┘  └─────┬───────┘  │HTTP API  │ │
│        │                │          │MCP Server│ │
│        └────────────────┼──────────┴────┬─────┘ │
│                         │               │       │
│          ┌──────────────┴───────────────┴────┐  │
│          │         AI Agent Node             │  │
│          │       (Reasoning Engine)          │  │
│          │  • Agentic Loop Processing        │  │
│          │  • Context & Memory Management    │  │
│          │  • Tool Selection & Execution     │  │
│          └───────────────────────────────────┘  │
└─────────────────────────────────────────────────┘

Step 1: Connect LLM Provider

Interactive demonstration of LLM provider connection setup in Axellero Studio

Configure OpenAI LLM Client Connection

1. Open Axellero Studio → Integrations section
2. Click Create New Connection → Select Open AI LLM Client from dropdown
3. Configure settings:

API Key Setup

1. Visit OpenAI Platform
2. Sign in and navigate to API Keys section
3. Click Create new secret key
4. Copy the key (starts with sk-) and store securely

Make sure there is no error after Connector creation

For detailed LLM provider configuration, see Connector documentation.

Step 2: Build Agent Architecture

Create and Configure a New AI Agent

Create New AgentFlow

Follow these steps to create and configure your AI agent:

1. Start agent creation - Click the "Create new" button to launch your first AI Agent setup
2. Enter agent code - Provide your custom code to define how your new AI Agent will work
3. Create the agent - Click "Create" to add your new AI Agent and begin leveraging intelligent automation
4. Open agent editor - Double click on the newly created "Axellero Docs AI Agent" in the agents list

Add AI Agent Node to Canvas

1. Add AI Agent node - Click on an outgoing link anchor on the Start node, drag and click on any point on the canvas to add an AI Agent node
2. Select node type - Choose "AI Agent node" from the Node Selector and add it to your AgentFlow to power up your project with AI capabilities
3. Open configuration - Click on the AI Agent node to open the agent's configuration panel and fine-tune its settings

Configure AI Agent Settings

1. Select LLM Model - Choose an LLM Model from your available Connector (if you have only one connector, it's automatically set)
2. Set System Message - Configure a System Message to guide your AI Agent's responses and tone
3. Access full configuration - Click the "Full view" button to expand the configuration panel of the AI Agent node
4. Edit System Prompt - Customize the System Prompt to fine-tune your AI Agent's communication style
5. Close configuration - Close the "Full view" when done to keep your workspace organized

System Prompt Configuration

You are an expert AI assistant specialized in Axellero platform documentation. Your role is to help users understand and implement Axellero features effectively.

## Your Expertise
- Axellero platform architecture and capabilities
- AI agent development and configuration
- Workflow design and automation
- Integration patterns and best practices

## Response Guidelines
- Always provide accurate, up-to-date information based on your knowledge base
- Include relevant code examples when helpful
- Reference specific documentation sections for detailed information
- If you don't know something, acknowledge limitations and suggest resources

## Tools Available
- Search documentation for specific topics
- Extract code examples and configurations
- Find best practices for specific use cases

Always maintain a helpful, professional tone while being thorough and technically accurate.

Configure Knowledge Base

Add Documents to Your AI Agent's Knowledge Base

Follow these steps to add documents that will power your agent with relevant knowledge:

1. Navigate to AI Agents - Select AI Agents to access and manage your tailored assistants
2. Switch to Knowledge Base tab - Click the Knowledge Base tab to enrich your agents with key resources
3. Select your knowledge base - Choose your knowledge base (e.g., "Axellero Docs") to view and manage its content
4. Add sources - Click "Add sources" to power your agent with essential documents or information
5. Browse for files - Use the browse link to find and select files from your device to add as knowledge sources
6. Upload documents - Hit "Upload" to instantly add your documents and expand your agent's expertise

Connect Knowledge Base to Agent Node

Connect an AI Agent to a Knowledge Base

Follow these steps to connect your knowledge base to the AI agent:

1. Add Knowledge Base node - Click and drag from the Knowledge Base outgoing anchor point to add a Knowledge Base node to your AgentFlow
2. Add to canvas - Add the Knowledge Base node to your AgentFlow for enhanced information access
3. Configure the node - Select the knowledge base node to view or organize its configuration
4. Choose knowledge base - Select your preferred Knowledge Base from the dropdown to tailor your agent's expertise
5. Confirm selection - Click to confirm your Knowledge Base selection or update details for accuracy
6. Add custom parameters - Add custom parameters to fine-tune how your agent connects to the knowledge base
7. Set result limit - Adjust the Limit setting to control how many results your agent retrieves
8. Configure score threshold - Set your preferred ScoreThreshold to fine-tune response relevance and accuracy
9. Update system prompt - Edit the agent's system prompt to include the newly added knowledge base

Knowledge Base Configuration Settings

For detailed knowledge base setup, see AI Knowledge nodes documentation.

Step 3: Test Agent Functionality

Test an AI Agent Workflow with Preview and Chat

Launch Chat Sandbox

1. Open Chat Sandbox - Press the "Play" button to open the built-in Chat Sandbox where you can test the behavior of your AI Agent by chatting with it
2. Start new conversation - Click "New Chat" to start a fresh conversation with your AI agent
3. Test agent responses - Type your message to interact with the AI agent and test its responses

Basic Functionality Tests

Test 1: Basic Context Variable Query

Input: How do I access the current user's information in a workflow? Show me the specific context variables I can use.
Expected: Agent explains workflow context variables like {{ctx.user.id}}, {{ctx.user.email}} with examples

Test 2: Advanced Context Usage

Input: I need to pass data between workflow nodes and access session information. What context variables are available for data flow and user session management?
Expected: Agent covers data passing context variables, session management, and inter-node communication

Test 3: Specific Implementation Help

Input: Can you show me examples of how to use workflow context variables in conditional logic or data transformations? I need both the syntax and real use cases.
Expected: Agent provides code examples showing context variable usage in conditions and transformations

Validation Checklist

• Agent responds within 3 seconds for basic queries
• Returns relevant documentation content from knowledge base
• Answers are well-structured and helpful
• No errors appear in execution logs

For comprehensive testing strategies, explore the agentflow builder in Axellero Studio.

Step 4: Deploy to Production

Deploy a New Version of an AI Agent

Deploy from AgentFlow Builder

Follow these steps to deploy your AI agent to production:

1. Start deployment - Click "Deploy" to begin shipping your latest AI agent improvements
2. Configure access control - Control who can access your agent by toggling "Anonymous launch" on or off
3. Set security options - Boost agent security by switching the "Runtime token" feature as needed
4. Select version type - Choose "Patch" to release quick improvements without major changes
5. Add description - Provide a concise description of your latest updates for full transparency
6. Deploy to production - Click "Deploy" to instantly publish your enhanced AI agent to production

Deployment Configuration Options

For detailed deployment options, see the Operate documentation (Coming Soon).

What You've Built

Your Axellero Docs Agent now has:

• AI Reasoning via GPT-4 connection
• Documentation Knowledge through RAG integration
• Production Deployment with version control
• Conversation Memory for context retention
• Security Controls via access and token management

Next Steps

Additional Resources

• Core Concepts: AI Agents and multi-agent patterns
• Studio Reference: AgentFlow Builder interface documentation
• Node Specifications: AI Agent Nodes technical details
• Community Support: Axellero community forums